Mike Schroepfer to Amodei: be specific, not alarmist, on AI risk

The back-and-forth comes after weeks of AI warnings from Amodei and OpenAI CEO Sam Altman, and against a backdrop of mounting calls for stricter control of AI. This week, tech leaders gathered at the White House, where President Trump and the CEOs of the most powerful AI companies signed a set of voluntary safety standards on Tuesday. Trump described the standards as "almost like a constitution, in a way." Among those present were Elon Musk, Mark Zuckerberg, Amodei, Jensen Huang, Greg Brockman, and Sundar Pichai.

Schroepfer's critique is less about whether the risks are real than about how they are discussed. He argues that sweeping warnings make it hard for anyone to engage, and that the conversation should instead be specific, consistent, and focused on concrete areas like cybersecurity and bio threats, plus the question of whether the rewards are worth the risks. The reported side conversation at the White House lunch suggests his view is not isolated among peers.

The stakes may hinge on whether Amodei and other AI leaders move toward that more specific framing, since the voluntary standards signed this week are not legally binding and the push for stricter control is still building. If the discussion stays at the level of big absolutes, it may be harder for policymakers and the public to act; if it becomes more concrete, the same warnings could carry more weight.